ABSTRACT
Background and Objective Accurate measurements of peripheral blood oxygen saturation (SpO2) are important in evaluating sleep patients with frequent desaturations due to pathological breathing events. This project compared synchronised SpO2 measurements at the finger, forehead and toe of patients undergoing Type 1 polysomnography (PSG) to evaluate potential SpO2 variability across the sites.
Methods Pulse oximetry SpO2 measurements were simultaneously and continuously recorded for 41 sleep patients at the finger, forehead and toe, and synchronised with PSG data. Recordings were scored for desaturations of ≤ 3% (peak to trough, lasting ≤10 seconds), signal dropouts, and artefact occurrences. Forehead and toe SpO2 measurements were compared against the finger as the standard PSG oximetry site.
Results Differences between anatomical sites for mean SpO2, mean number of SpO2 desaturations per hour, and time spent below an SpO2 level of 95% during total sleep time were significant (P < 0.01). The forehead pulse oximeter had the highest mean SpO2, least number of SpO2 desaturations per hour, and experienced the least number of artefact occurrences. Dropouts were lowest for the forehead and toe pulse oximeters.
Conclusion Differences between SpO2 measurements, dropouts and artefact occurrences at the finger, forehead and toe may have diagnostic and prognostic implications for sleep patients. The differences in SpO2 measurements may be attributed to variability in perfusion of the extremities and core during sudden oscillating blood pressure changes associated with breathing events. Further research is required to determine which anatomical site correlates closest to arterial oxygenation for pulse oximetry in sleep patients.
Study rationale An absence of research investigating anatomical site location for pulse oximetry during overnight polysomnography exists. Our study was performed to fill this gap, as accurate pulse oximetry measurements are key for the diagnosis, treatment and monitoring of sleep patients; a patient cohort where SpO2 desaturations are recurrent.
Study impact Our findings demonstrate there are significant differences between finger, forehead and toe pulse oximetry measurements, particularly SpO2 desaturations per hour, which may have diagnostic and clinical implications. This research is applicable and important to not only sleep physicians and scientists, but also other disciplines where continuous SpO2 monitoring is required.
INTRODUCTION
Pulse oximetry is a non-invasive technique used in sleep diagnostics to assess patient blood oxygenation (SpO2) levels and for detecting hypoxaemia 1-4. A sleep study, also known as polysomnography (PSG), is used to diagnose obstructive sleep apnoea (OSA) and other sleep disorders. Patients undergoing PSG for the investigation of sleep disordered breathing often demonstrate significant and numerous blood oxygen desaturations that are a hallmark characteristic of OSA 2, 3, 5. PSG provides a comprehensive multi-parametric recording of the physiological changes that occur during sleep and detects pathological breathing events 2,3. Alternatively, an overnight oximetry study either at home or in hospital can be conducted as a surrogate investigation or initial screening process to assess sleep disorders. Both types of sleep studies require continuous and accurate monitoring of patient SpO2 levels 2,3.
Pulse oximetry is limited by factors such as peripheral perfusion, motion artefact and vasoconstriction 1, 6-10, which may be more prevalent when SpO2 is measured at different anatomical locations. Previous studies have reported that forehead pulse oximeters provided a more accurate and consistent SpO2 reading than finger pulse oximeters in patients undergoing post-anaesthesia care, ambulance transportation and respiratory assessment 11-15. Anatomical characteristics of the forehead, such as a higher level of tissue perfusion, fewer occurrences of vasoconstriction, and less risk of motion artefacts, may have contributed to the observed difference in SpO2 measurements 11-15. Two studies reported that SpO2 measurements with earlobe pulse oximeters in patients with chronic obstructive pulmonary disease (COPD) overestimated arterial blood oxygen saturation (SaO2) to such an extent that the earlobe was deemed inappropriate for evaluating oxygen administration and overall clinical decision-making 16,17.
Limited research exists comparing simultaneous SpO2 measurements on patients undergoing PSG using pulse oximeters at different anatomical sites. The failure of a pulse oximeter to detect an oxygen desaturation during a PSG means that the corresponding breathing event may not be scored, and the overall severity of the patient’s clinical condition may be underestimated or treatment under-titrated 2,3. The aim of this cross-sectional study was to compare simultaneous, synchronised SpO2 measurements obtained with pulse oximeters located at three anatomical sites (finger, forehead and toe) on patients undergoing Type 1 PSG. The mean SpO2, number and magnitude of SpO2 desaturations per hour, number of signal dropouts, and occurrence of artefactual readings were compared for pulse oximetry measurements at the three anatomical sites.
METHODS
Ethical approval was granted by The Prince Charles Hospital Human Research Ethics Committee (HREC/18/QPCH/179), with subsequent authorisation to conduct the project at the Sleep Assessment Unit at the Sunshine Coast University Hospital (SCUH) (QSC/40666). Patients undergoing Type 1 PSG were recruited from August 2018 to December 2018 by non-probability convenience sampling. Exclusion criteria included: any form of skin infection at the three anatomical sites; impaired nail quality; any medical condition known to affect pulse oximeter accuracy; and an inability to participate in Type 1 PSG. Patients recruited for the project completed a self-reported, medical history checklist consisting of 38 common medical conditions.
Patients had a pulse oximeter attached at their finger, forehead and toe, with the photo-emitter and detector placed as per each manufacturer’s specifications. A NONIN pulse oximeter with soft-tip finger probe was placed on the middle finger as per standard practice for Type 1 PSG. A Masimo pulse oximeter integrated into a Sensor 92 on a Radiometer TCM4 TOSCA monitoring system was used on the forehead and set to 43°C. A Masimo Radical 7 pulse oximeter with disposable probe was placed on the best fitting toe. Oximetry outputs were recorded at a sample rate of 16Hz. The three oximeters were recorded simultaneously in conjunction with other routine electrode channels and synchronized with other PSG parameters. All electrodes were monitored continuously overnight by a sleep scientist in the Sleep Assessment Unit. Pulse oximetry data was scored according to strict rules (Fig. 1) to differentiate the commencement and termination of SpO2 desaturation events. To avoid scorer bias, pulse oximeter data was scored without the anatomical site label. Each patient’s full PSG was scored according to the American Academy of Sleep Medicine (AASM) manual in conjunction with Australasian Sleep Technologists Association (ASTA) and Australasian Sleep Association (ASA) commentary 2,3.
De-identified patient data including PSG results, pulse oximetry measurements, age, gender, height, weight and body mass index (BMI) were transferred to SPSS® (Version 24, IBM) for statistical analysis. Descriptive statistics were used to quality check and summarise the data, and reported as mean (SD). Parametric tests were used for inference due to the sample size (n=41). The assumptions for all statistical tests performed were met. Levene’s test for equality of variances was undertaken before conducting all unpaired t-tests. A Repeated Measures Analysis of Variance (RMANOVA) was used to determine significant differences between the three pulse oximeters for all SpO2 measurements. The relationship between oximeter values were determined using Pearson’s correlation coefficient (r). Bland-Altman plots were used to complement the correlation analysis, and to show the level of agreement between the oximeters at the different anatomical sites, using the finger as the standard PSG oximetry site 18, 19. Statistical significance was chosen at P < 0.05.
RESULTS
Table 1 summarises the demographic attributes of the Type 1 PSG patients (n=41) recruited for the project. The patients underwent one of the following: diagnostic study (39%, n=16), split diagnostic/Continuous Positive Airway Pressure (CPAP) study (19.5%, n=8), CPAP titration study (36.6%, n=15) or bilevel positive airway pressure titration study (4.9%, n=2). Most patients (91%, n=30) had sleep disordered breathing, including a mild OSA (42.4%, n=14) or moderate OSA (24.2%, n=8). Half the patients (51.2%, n=21) had four or more medical conditions, of which 17% had ten or more conditions. Many patients reported a cardiac history (58.5%, n=24) or respiratory condition (40%, n=17). Other reported conditions were anxiety (43.9%, n=18), depression (36.6%, n=15), gastric reflux (41.5%, n=17) and diabetes (29.3%, n=12).
Mean SpO2 during total sleep time at the forehead was significantly higher than toe oximetry and finger oximetry measurements (RMANOVA: F (1.245, 48.551) = 38.782, P < 0.01; Table 2). The percentage difference in both mean magnitude of SpO2 desaturations and largest SpO2 desaturation during total sleep time was smallest between the finger and forehead pulse oximeters. While there was no significant difference between the three pulse oximeters for the largest SpO2 desaturation during total sleep time (RMANOVA: F (1.333, 49.311) = 2.972, P = 0.08), the mean number of SpO2 desaturations detected per hour were significantly different (RMANOVA: F (1.427, 57.066) = 14.891, P < 0.01, Table 2). The finger pulse oximeter detected the most SpO2 desaturations per hour and the forehead pulse oximeter detected the least number. A high degree of variability existed with all pulse oximeters for the detection of the mean number of SpO2 desaturations per hour.
The mean number of SpO2 desaturations per hour during total sleep time showed a strongly positive and significant relationship between the finger and forehead pulse oximeters (R2 = 0.937, P < 0.01, Fig. 2A) and the finger and toe pulse oximeters (R2 = 0.942, P < 0.01, Fig. 2C). A Bland-Altman plot confirmed that the mean number of SpO2 desaturations detected per hour by the finger pulse oximeter was higher than both the forehead and toe oximeters, with wide limits of agreement of -14.56 to 29.42% (Fig. 2B) and -11.89 to 22.20% (Fig. 2D), respectively. The mean number of SpO2 desaturations per hour were more likely to fall outside the limits of agreement when values were greater than 20 desaturations per hour.
A significant difference was observed between the three pulse oximeters for the time spent at an SpO2 below 95% during total sleep time for the 41 patients (RMANOVA: F (1.178, 45.933) = 24.430, P < 0.01, Table 3). The finger pulse oximeter spent the greatest time below an SpO2 of 95% and the forehead pulse oximeter the least time. For one patient, the finger pulse oximeter spent over 3 hours longer below an SpO2 level of 95% than the forehead and toe pulse oximeter. While SpO2 dropped below 88% during total sleep time for 11 patients (Table 3), the difference between pulse oximeters was not significant (RMANOVA: F (1.051, 10.506) = 3.144, P = 0.104).
Table 4 shows the finger pulse oximeter had a significantly higher mean number of signal dropouts per patient (RMANOVA: F (1.086, 41.336) = 12.544, P < 0.01) and dropouts per hour (F (1.095, 42.718) = 12.804, P < 0.01) than the forehead and toe pulse oximeters during total report time. Furthermore, the forehead pulse oximeter had significantly lower mean occurrences of artefact (RMANOVA: F (1.218, 47.487) = 4.594, P = 0.030) and rate of artefact per hour (RMANOVA: F (1.237, 48.235) = 4.969, P = 0.024) than the finger and toe pulse oximeters during total report time.
DISCUSSION
Our study found differences between SpO2 values measured with pulse oximeters positioned at the finger, forehead and toe of patients undergoing Type 1 PSG, which is consistent with previous studies of patients in post anaesthesia care, intensive care units, ambulatory transportation, and with chronic obstructive pulmonary disease 11-17, 20-26. The finger pulse oximeter, which is the standard PSG oximetry site, recorded the lowest mean SpO2 and highest mean number of SpO2 desaturations per hour during total sleep time. However, the finger pulse oximeter also experienced the highest number of dropouts per patient and per hour as compared to the forehead and toe pulse oximeters. Artefact occurrences were highest for both the finger and toe pulse oximeters. Such differences between oximeters for the continuous monitoring of SpO2 during total sleep time may have diagnostic and prognostic implications for patients undergoing PSG, including the assessment of OSA severity with the AHI and other oxygen saturation indices 2-4.
Several studies have reported that forehead oximetry recorded a higher mean SpO2 and had better agreement with fewer outliers than finger oximetry when compared to SaO2 for respiratory and intensive care patients 11, 12, 15. In contrast, Sohila et al. found the forehead oximeter recorded a lower mean SpO2 than finger, toe and earlobe oximeters 26. While our study showed differences for mean SpO2 across total sleep time between the three oximetry sites, these differences were within acceptable limits of agreement (± 2%) set by manufacturers for their pulse oximeters 6, 27 and are unlikely to be clinically significant.
Patients with OSA experience frequent blood oxygen saturations associated with pathological breathing events 5. We found poor agreement between the three oximeter sites when the frequency of SpO2 desaturations per hour was >20, with some SpO2 values falling outside the wide limits of agreement (Figs. 2A & 2C). This could influence patient diagnosis and classification of OSA severity, and result in suboptimal treatment, such as the prescription of inadequate CPAP pressures 2, 3. For instance, the classification of OSA severity for one of the patients in our study was graded as severe with finger oximetry (55 SpO2 desaturations per hr), moderate with toe oximetry (21 SpO2 desaturations per hr), and mild with forehead oximetry (14 SpO2 desaturations per hr). Differences in frequency of SpO2 desaturations detected per hour by each oximeter is concerning for home sleep apnoea testing and overnight oximetry screens that depend on oximetry measurements alone and lack informative supporting PSG parameters35.
The vigorous cycle of sudden oscilating blood pressure and perfusion changes due to obstructive respiratory events may have different effects on oximetry readings at the core and extremities. Vasoconstriction and vasodilation can cause oximeters to overestimate or underestimate SaO2 in healthy volunteers 19, 10, 28, 29. Furthermore, the degree of hypoxaemia experienced during obstructive respiratory events may be less severe at the forehead than extremities because of the protective cerebrovascular autoregulation mechanism 30-32, thereby explaining why the forehead oximeter detected the least number of SpO2 desaturations per hour. The clinical significance of our findings requires further investigation.
The finger oximeter spent the most time below an SpO2 of 95% (Table 3), with one patient recording over 3-hours longer below an SpO2 of 95% than the forehead and toe oximeters. Since the baseline SpO2 was approximately 95% for some patients, small variations in SpO2 above and below this clinical limit are likely to have contributed to the poor agreement between oximeters. In contrast, the difference in time spent below 88% rarely exceeded 10 minutes, suggesting good agreement between oximeters at this clinical limit. While this difference is insignificant during a total sleep period, it could be problematic in other clinical settings such as ICU where measurement time of SpO2 is critical 33, 34.
Since SpO2 measurements are crucial for apnoea and hypopnea scoring in patients undergoing PSG, signal quality of pulse oximeters at each anatomical site is imperative 2, 3, 26,. In our study, the forehead oximeter had better signal stability and reliability than the finger and toe oximeters, with only ten dropouts and six artefact occurrences being recorded across 280 hours of sleep data for 41 patients (Table 4). Similar findings were reported for forehead oximetry in patients undergoing ambulance transportation and postanaesthesia care 13, 14. This suggests that signals from pulse oximeters placed on the extremities are less stable and reliable than those placed more central to the core. Increased risk of motion artefact, vasoconstriction, and decreased perfusion at the extremities, all of which can weaken the photoplethysmographic signal 7, 9, 10, may contribute to oximetry instability at the three anatomical sites selected for our study.
LIMITATIONS
Our research study had several limitations. First, comparing the pulse oximetry SpO2 measurements against arterial blood gas (ABG) measurements to ascertain which pulse oximeter recorded values closest to SaO2 values was not possible. ABG samples taken from patients undergoing PSG are impractical because they are invasive and painful, and not recommended for standard clinic practice 2, 3. Secondly, the finger pulse oximeter was manufactured by a different company to the oximeters attached to patients’ foreheads and toes. However, pulse oximeters utilise similar theoretical principles and algorithms and should not vary substantially to influence SpO2 measurements. Finally, our cross-sectional study included a small sample of sleep patients with a range of medical conditions. The study exclusion criteria sought to minimise confounders of SpO2 measurements at the three anatomical sites. Nevertheless, SpO2 measurements at the anatomical sites may have been affected by individual pathologies in unknown ways, which requires further investigation.
CONCLUSION
Our study has compared SpO2 measurements using pulse oximetry at three anatomical sites (finger, forehead, toe) on patients undergoing Type 1 PSG. The forehead pulse oximeter recorded the highest mean SpO2 level, detected the least SpO2 desaturations, and had the most stable and reliable signal in relation to dropouts and artefact occurrences. The differences in SpO2 measurements from the three oximeter sites may be attributed to variability in perfusion between the extremities and core during sudden oscillating blood pressure changes associated with obstructive breathing events. We recommend that careful consideration be given to the interpretation of SpO2 measurements from different oximetry sites for the diagnosis and treatment of patients undergoing PSG. Clearly, further research is required to determine with confidence which pulse oximetry site most accurately reflects changes in SaO2 levels for patients undergoing sleep studies.
Data Availability
Patient data supporting the findings of this project are available from the Sunshine Coast Hospital and Health Service (SCHHS). Restrictions apply to the availability of these data, which were used under license for this project. Data are available upon request.
DATA AVAILABILITY STATEMENT
Patient data supporting the findings of this project are available from the Sunshine Coast Hospital and Health Service (SCHHS). Restrictions apply to the availability of these data, which were used under license for this project. Data are available from Chris Brown with permission of the SCHHS.
ACKNOWLEDGEMENT
The authors acknowledge the SCUH Sleep Assessment Unit and SCHHS for the use of their clinical facilities and assistance from their sleep scientists. This project received no external funding.
Footnotes
This research project was performed within the Sleep Assessment Unit at the Sunshine Coast University Hospital. All authors have seen and approved the manuscript and agree with its submission to MedRxiv. This project received no external funding, and all authors have no conflicts of interest to declare.
ABBREVIATIONS
- AASM
- American Academy of Sleep Medicine
- ABG
- Arterial Blood Gas
- AHI
- Apnoea Hypopnea Index
- ATSA
- Australasian Sleep Technologists Association
- ASA
- Australasian Sleep Association
- BMI
- Body Mass Index
- COPD
- Chronic Obstructive Pulmonary Disease
- CPAP
- Continuous Positive Airway Pressure
- ICU
- Intensive Care Unit
- OSA
- Obstructive Sleep Apnoea PSG: Polysomnography
- SaO2
- Arterial blood oxygen saturation
- SpO2
- Peripheral blood oxygen saturation